Research On Human-Like Autonomous Lane-Changing Algorithm On Highways

As a common driving behavior on highways,lane-changing has many effects on traffic flow.How to achieve safe and reasonable autonomous lane-changing is a hot research topic in the field of autonomous driving.Based on the above background,and in order to make the autonomous lane-changing action closer to human driving behaviors,and allow passengers to obtain better acceptance and satisfaction,this article relies on the Ministry of Science and Technology’s key special project "Research and Demonstration Operation of Key Technologies for Electric Autonomous Vehicles",carried out the research on the human-like autonomous lane-changing algorithm under the highways.First,set up a lane-changing data collection platform to record the driver’s lane-changing behavior data,and extract lane-changing characteristic parameters that reflect the driver’s behavior.Then,study the human-like autonomous lane-changing behavior trigger mechanism,and use polynomial curves to achieve lanechanging trajectory planning.Finally,based on the preview-following theory,the trajectory tracking controller was designed,and the Car Sim & Simulink co-simulation platform and the real-vehicle test platform were built,the algorithm was tested under typical working conditions.In order to analyze the driver’s lane-changing behavior,this paper builds a lanechanging data collection platform to record the driver’s lane-changing data under a variety of typical working conditions.Then,based on the test data,this paper analyzes the vehicle motion state when changing lanes at a single speed and multiple speeds,and explores the changing laws of vehicle kinematics and dynamic parameters.Finally,the human-like characteristic parameters that can reflect the driver’s lane-changing behavior are extracted,and provide human-like data reference for the subsequent research on the trigger mechanism of lane-changing behavior and trajectory planning methods.In order to simulate the driver’s lane-changing decision-making behavior,this paper studies the trigger mechanism of human-like autonomous lane-changing behavior around the cause of lane-changing intention,target lane selection method and lanechanging safety judgment strategy.First,define the lane tolerance variable to describe the cause of the lane-changing intention.Secondly,consider the speed advantage to select the target lane.Then,establish an improved minimum safe spacing model to judge the safety of lane-changing.Finally make the driving decision for lane-changing or lane-keeping.In order to plan the lane-changing trajectory,the potential field method is used to construct a safe area around the vehicle,taking the dynamic changes of the driving environment into account,an improved potential field coefficient is introduced to make the risk influence range of the environmental vehicle varies with the change of relative speed.Then,based on the human-like lane change time and the fifth-order polynomial curve,the candidate trajectory in the safe area is planned.Finally,the evaluation function considering safety,acceptance and efficiency is designed,and it is used to make the reasonable choice of lane-changing trajectory.In order to achieve a good tracking of the target trajectory,this paper designs a trajectory tracking controller based on the two-degree-of-freedom model and previewfollowing theory.First,the vehicle-road model is established,and the preview point search method on the target trajectory is proposed.Then,in view of the limitation of the fixed preview time when driving at high speed,the adaptive preview time method based on the curvature of the target trajectory is proposed.Finally,considering the lateral deviation and lateral acceleration deviation,a feedforward-feedback trajectory tracking controller is designed.Finally,in order to verify the rationality and effectiveness of the algorithm,a simulation platform based on Car Sim & Simulink and a real vehicle test platform based on Haval H7 were built,and typical driving conditions were designed to verify the algorithm function.Simulation and actual vehicle test results show that the decisionmaking method based on human-like characteristic parameters can make human-like lane-changing behavior decisions and reasonable trajectory planning for a variety of driving conditions,and the feedforward-feedback controller with adaptive preview time can achieve good tracking of the target trajectory.